Skip to main content

Full text of "Of a certain age : women and age discrepant marriages"

See other formats


OF A CERTAIN AGE: 

WOMEN AND AGE DISCREPANT MARRIAGES 



By 

SYLVIA MARION CARLEY 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT 
OF THE REQUIREMENTS FOR THE DEGREE OF 
DOCTOR OF PHILOSOPHY 



UNIVERSITY OF FLORIDA 



1988 



• 5WKITY OF FLORIDA LIBRARIES 



Copyright 1988 
by 

Sylvia Marion Carley 



Dedicated to Henry, Ivana, Mom, Gramps, my support 
group members, and especially to the memory 

of my Dad 



ACKNOWLEDGEMENTS 



I wish to express my earnest appreciation to Dr. 
Constance Shehan, who took time to share with me her 
knowledge and skills. I would also like to thank Dr. Felix 
M. Berardo, who never lost the faith and for being there to 
provide guidance. In addition, I must express gratitude to 
Dr. Hernan Vera for sharing with me his data and 
explanations of how best to obtain the information for this 
entire project. Finally, I wish to acknowledge my special 
teachers and committee members Dr. Lee Crandall and Dr. Otto 
VonMering for their support and efforts. 

This project's completion represents the collaborative 
efforts of many of my family members, colleagues, and 
friends. Each of them assisted in their own special way 
but I want to thank them for being so understanding when I 
could not participate, could not come or could not call. I 
must, however, acknowledge the prayers of Gramps, aunties 
Johnnye Mai, and Florence, the proofing of Altamese, the 
discussions with Candy, the computer assistance of Bob and 
last but not least the librarian assistance provided by Ray 
Jones . 

Finally, I am deeply grateful to Henry , my husband, 
for being there when I needed him most and for enduring with 
patience. 



iv 



TABLE OF CONTENTS 



Page 



ACKNOWLEDGEMENTS iv 

LIST OF TABLES vii 

ABSTRACT ix 

CHAPTER 

ONE INTRODUCTION 1 

Purpose and Implications of the Study 4 

Theoretical Implications 4 

TWO REVIEW OF MATE SELECTION THEORIES 6 

Complementary Needs V 

Consensus Theory 9 

Stimulus Value Role Theory 10 

Filter Theory 11 

THREE HOMOGAMY AND HETEROGAMY IN MATE SELECTION 15 

Homogamy 15 

Age Homogamy IV 

Age Heterogamy 2 0 

Women Who Marry Older Men 25 

Women Who Marry Younger Men 28 

Summary 33 

FOUR SOCIOCULTURAL DETERMINANTS OF AGE DISCREPANCY 

IN MARRIAGES 34 

Racial and Ethnic Differences 34 

Education/Social Class 37 

Marital History 39 

Sex Ratio 41 

FIVE METHODOLOGY 4 5 

Operational Definition 45 

Data 4 6 



v 



Page 



Variables 47 

Analytical Procedure 50 

SIX FINDINGS AND DISCUSSION 54 

Descriptive Characteristics 54 

Describing the Subpopulations 58 

Logistic Regression Results 64 

Summary 8 6 

SEVEN CONCLUSIONS 94 

Research Focus 95 

Issues of Concern and Ideas for Future 

Research 98 

REFERENCES 105 

BIOGRAPHICAL SKETCH 116 



vi 



LIST OF TABLES 



Table Page 



1. DESCRIPTION OF VARIABLES IN THE ANALYSIS 48 

2. CHARACTERISTICS OF AGE HOMOGAMOUS AND AGE HETER- 

OGAMOUS: ALL COUPLES COMBINED 55 

3. MEANS AND STANDARD DEVIATIONS OF VARIABLES USED 

IN THE ANALYSIS, ALL CASES 57 

4. CHARACTERISTICS OF BLACK WOMEN IN AGE HOMOGAMOUS 

AND AGE HETEROGAMOUS MARRIAGES 59 

5. CHARACTERISTICS OF WHITE WOMEN IN AGE HOMOGAMOUS 

AND AGE HETEROGAMOUS COUPLES 62 

6. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WOMEN'S 

LIKELIHOOD OF BEING IN AN AGE DISCREPANT 

MARRIAGE: BLACK AND WHITE WOMEN, COMBINED 66 

7. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WOMEN'S 

LIKELIHOOD OF BEING IN AN AGE DISCREPANT 

MARRIAGE IN WHICH THE WIFE IS OLDER BY 5 OR 

MORE YEARS 69 

8. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WOMEN'S 

LIKELIHOOD OF BEING IN AN AGE DISCREPANT 

MARRIAGE IN WHICH THE WIFE IS YOUNGER BY 10 OR 

MORE YEARS 71 

9. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON BLACK 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 
MARRIAGE 74 

10. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON BLACK 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 
MARRIAGE IN WHICH THE WIFE IS OLDER BY 5 MORE 
OR YEARS 76 

11. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON BLACK 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 

MARRIAGE IN WHICH THE WIFE IS YOUNGER BY 10 

OR MORE YEARS 78 



vii 



Table 



Page 



12. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WHITE 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 
MARRIAGE 80 

13. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WHITE 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 

MARRIAGE IN WHICH THE WIFE IS OLDER BY 5 OR 

MORE YEARS 8 3 

14. EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WHITE 

WOMEN'S LIKELIHOOD OF BEING IN AN AGE DISCREPANT 
MARRIAGE IN WHICH THE WIFE IS YOUNGER BY 10 OR 
MORE YEARS 8 5 



viii 



Abstract of Dissertation Presented to the Graduate School 
of the University of Florida in Partial Fulfillment of the 
Requirements for the Degree of Doctor of Philosophy 

OF A CERTAIN AGE: 

WOMEN IN AGE DISCREPANT MARRIAGES 

by 

Sylvia Marion Carley 
August 1988 

Chairman: Felix Berardo 
CoChairman: Constance Shehan 
Major Department: Sociology 

Just as there are norms about the age at which women 
and men should marry, there are norms about age differences 
between spouses. The established age differential pattern 
is approximately 2 to 3 years between wives and husbands. 
This pattern of "constrained heterogamy" has been accepted 
in the American society as the "norm" in mate selection. 
The purpose of this research is to identify the factors 
associated with women's likelihood of marrying outside this 
normative age pattern. To ascertain which women who far 
exceeded the "constrained heterogamous" pattern in mate 
selection, age discrepancy was defined as women who are 5 or 
more years older than their spouses or 10 or more years 
younger than their spouses. 

A sample of 46,844 married women from the 1980 United 
States Census is examined to determine, first, the likeli- 
hood of women participating in age discrepant marriage, 
second, the racial differences in the determinants of parti- 
cipation. 



ix 



Results indicate women in their first marriage are 
less likely to participate in age discrepant marriages. If 
women are married in educationally, racially or ethnically 
heterogamous marriages they are more likely to participate 
in age discrepant marriages. Having a Spanish ethnic 
background decreases the likelihood for black women to 
participate in age discrepant marriages; however, it 
increases the likelihood for white women. The 
sociodemographic variables were not as predictive with wife 
older marriages as they were with wife younger. Although 
this research has provided some knowledge of the predictors 
of age discrepant marriages, additional studies are needed 
to offer more information on the role expectations of the 
husbands and wives and to assess the prevalence in the 
society. 



x 



CHAPTER ONE 
INTRODUCTION 



One of the most significant decisions a person faces is 
whether or not to marry. If one elects to marry, the choice 
of a particular mate becomes the next crucial decision. 
Mate selection is seldom a random activity. Moreover, one 
of the qualifiers or disqualif iers of a potential partner is 
age. 

Age is not a neutral or simple variable. It is 
socially defined and enormously consequential in the eyes 
of members of the society. Modell (1980:211), in his analy- 
sis of American marriage timing states, "Ours (American) is 
an age stratified society, more precisely, a gender strati- 
fied society. Privileges and obligations are assigned in 
large part attendant upon chronological age". The United 
States is frequently characterized as an age conscious 
nation. Hence, it is not surprising that this awarenesss 
can also be seen in its norms governing mate selection and 
marriage . 

Most people in the United States marry for the first 
time during a relatively narrow age span and, hence, are 
generally quite similar in age to their spouses. Census of 
population data indicate that, on the average, American men 
marry for the first time at age 25 and women marry at age 



- 1 - 



- 2 - 



23 (U. S. Census Bureau, 1985). Consequently, this age 
similarity is in fact a type of age dissimilarity since the 
norm is actually one of age difference, with husbands ex- 
pected to be older than their wives. Thus, strictly 
speaking, the norm is a type of "constrained dissimilarity" 
(Vera, Berardo, and Berardo, 1985) . 

Why is there an overwhelming likelihood that females 
will be younger than males when they marry? Rossi (1986) 
has noted that the custom of younger wives marrying older 
husbands came about partly because a man could "father" a 
child for several decades longer than a woman could 
conceive. This particular motivation for entering age 
discrepant marriages will probably change now that families 
are smaller and the reproduction goals are achieved earlier 
in the life cycle. However, the normative expectation that 
the husband be older than his wife is generally accepted 
without hesitation by members of American society and will 
influence mate selection regardless of fertility concerns. 

These normative expectations have been reflected in 
much of the existing social science research literature on 
age discrepant marriages. In their review of the literature 
on such marriages Vera, Berardo, and Berardo (1985) identi- 
fied biases in several sociological research studies. For 
example, Blood and Wolfe (1960) studied satisfaction with 
standard of living among wives and concluded that age 
homogamous couples were the most satisfied and that 
satisfaction declined as the age gap increased in either 



-3- 



direction. Blood and Wolfe (1960:211-212) attributed this 

to unhealthy motivations and psychological needs of the 

husbands in the age discrepant marriages. For instance, in 

an analysis of 16 couples in which the husbands were four 

or more years younger, the authors concluded that 

Many of them marry older women as mother 
substitutes on whom they can depend for 
emotional security. . . . these over mothered 
husbands may duck out of home to find some 
satisfying companionship elsewhere. . . . 

Blood and Wolfe (1960:211-212) also explained the 

lower scores in marital satisfaction among the 16 wives: 

On occasions when a mother-wife does meet 
with frustration, she cannot turn to her 
"little boy" and expect him to nurture 
her. If she does, she is bound to be 
disappointed or even hurt, since his 
gregarious ego defends itself against 
its inadequacy by lashing back at her. 

On the other hand, the 37 husbands who were 11 or more 
years older than their wives were characterized by Blood and 
Wolfe (1960:163) as marrying a daughter image: Their pater- 
nalism tends to go beyond mere sympathy into direct advice 
about how the wife 

can solve problems. Such husbands play the role 
of "big strong men" on whom the wife depends 
for emotional support and decisions, but aren't 
likely to work beside her. They are good at doing 
things for their wives, not with them. 

Other authors, such as Derenski and Landsburg (1981: 

81) , argue that among the real reasons for the negative 

reaction against older women/younger men relationships is 

that these couples are seen as violating the incest taboos: 

While older women-younger men are not 
committing incest, the age difference between 



- 4 - 



them reminds people of the forbidden sexual 
relation between mothers and sons. Very 
strong feelings of social disapproval arise 
about this taboo because of its extreme 
cultural, historical, and psychological 
significance . 

These are only a few of the biases found in the empiri- 
cal literature on age discrepant unions. They illustrate 
that social scientists have not escaped the prejudice held 
against these couples and have not succeeded in explaining 
why individuals violate such strongly held beliefs. 

Purpose and Implications of the Study 

Because the existing research on age discrepant unions 
is limited by psychological reductionism, the purpose of 
this research is to identify the factors associated with a 
woman's likelihood of marrying outside the normative age 
patterns, which can lead to severe societal sanctions such 
as ostracism and isolation. In addition, this research will 
provide a descriptive analysis of the female participants in 
age discrepant marriages. Because census data clearly indi- 
cate a differential sex ratio by race, with a considerable 
shortage of black men in the "marriageable" ages, this 
research will compare black and white women's likelihood of 
entering these age discrepant marriages. 

Theoretical Implications 

This research has both theoretical and practical impli- 
cations. Theoretically, it will provide data concerning 



-5- 



norm violation with respect to homogamous mate selection. 
On the applied level, information derived from this study 
will provide some insight into potential expansion of the 
pool of eligible spouses, given recent reports concerning 
the difficulty well-educated women appear to be having in 
finding "suitable" spouses (Westoff and Goldman, 1984). In 
other words, patterns observed in these data may identify 
women who redefine "suitability" and expand their pool of 
eligibles by going outside the conventional age category. 

The remainder of this thesis is organized as follows: 
Chapter Two discusses mate selection theories. Chapter 
Three presents the generalizations of homogamy, age homog- 
amy, age heterogamy. Chapter Four discusses the previous 
research literature on the sociocultural determinants of age 
discrepant marriages and presents the hypotheses for the 
current research. Chapter Five provides the operational 
definition of age discrepant marriages and describes the 
source of the data and the methods used to analyze them. The 
analysis will review the patterns for all couples and then 
for black and white couples. Chapter Six presents the 
research findings, and Chapter Seven discusses and develops 
conclusions about the likelihood of female participation in 
age discrepant marriages. Since many of the studies on age 
discrepant marriages have not been empirically derived, this 
project will make a signficant contribution to the study of 
age as a factor in the understanding of mate selection. 



CHAPTER TWO 

REVIEW OF MATE SELECTION THEORIES 

Over two million American couples marry each year (U.S. 
Census, 1985). Although most couples believe "love," or 
perhaps "astrology," has brought them together, the more 
realistic mechanism is the filtering and sorting action of 
social structure. Indeed, all societies have mechanisms for 
controlling who gets married to whom. Cultural patterns are 
always evident in the process of mate selection (Murstein, 
1976) . In parts of the world, such as India and Turkey, 

arranged marriages in which prospective spouses have little 
or no decision-making power still occur. It has been re- 
ported that in Turkey, three-quarters of the marriages were 
arranged by the families or their designated intermediaries 
(Fox, 1975) . In many areas, the arrangement of marriages may 
be largely ceremonial, but couples are far from having the 
kind of free choice taken for granted in other nations. 
Young people in those countries may have a "veto" over 
parents' choices for them, but few would defy their parents 
by marrying someone of whom the parents disapproved. 

However, the high value placed on individualism and 
free will in western countries generates the notion that 
citizens operate within an open mate selection system. In 



- 6 - 



-7- 



fact, a number of normative and structural factors restrict 
the selection of a mate. Each society provides normative 
guidelines for those contemplating marriage and relies on a 
variety of informal mechanisms to ensure that the majority 
of marriages occur between appropriate groupings. According 
to Melville (1977), although the systems of arranged mar- 
riages and mate selection (based on attraction and free 
choice) seem radically different, bargaining is involved in 
both. For instance, Melville (1977:29) states, 

In all societies, whether it is 
acknowledged or not, the mate choice 
takes place within a marriage market. 

Like other markets, this one involves 
a commodity — an eligible male or 
female — to be exchanged at a certain 
price. The currency in the exchange 
consists of the socially valued attri- 
butes and skills of the two individuals' 
family background, economic position, 
education, and personal attributes 
such as age and beauty. Whatever the 
qualities are that the society considers 
valuable, each family or individual seeks 
as a mate the person who has the most of 
them. 

In this chapter, several major approaches of mate selec- 
tion will be discussed. These include (1) the complemen- 
tary needs theory, (2) the consensus theory, (3) the 
stimulus value role or (svr) theory, and (4) filter theory. 



Complementary Needs 



This perspective presumes that human behavior is ori- 
ented toward the gratification of needs. Important needs 
become organized in the personality and give pattern to 



- 8 - 



behavior. Winch (1958) , the leading proponent of this 
theory, has tended to limit his discussions to the attrac- 
tiveness of complementary personality needs. Thus, in mate 
selection, each person selects from within the field of 
eligibles that person who gives greatest promise of provid- 
ing the maximum need gratification. Furthermore, for the 
couples a significant aspect of being in love is the reali- 
zation that they are emotionally interdependent. The part- 
ners' need patterns, consequently, will be complementary 
rather than similar. Winch (1958) described two types of 
complementarity. In the first, known as Type I Complemen- 
tarity, a person with a high need in certain areas is at- 
tracted to a person with few needs in the same areas. Type 
II Complementarity occurs when people are attracted to each 
other because each has different needs that the other can 
satisfy. 

To test his hypothesis of mate selection, Winch (1958) 
used a sample of 25 undergraduate, childless couples at 
Northwestern University. Need interviews and case histo- 
ries were supplemented through the use of projective test- 
ing. Ratings of personality need were then based upon con- 
tent analysis of the need interview and case history mate- 
rials, summarized, and evaluated at conferences with one of 
the researchers. He suggested that the bulk of the evidence 
from these couples supports the hypothesis that mates tend 
to select each other on the basis of complementary needs 



-9- 



(Winch, 1958) . The results required numerous qualifica- 
tions, and the findings have been criticized by other 
researchers who have not been able to replicate Winch's 
study. 

Consensus Theory 

The consensus theory suggests that (1) persons with 
similar backgrounds learn similar values; (2) the interac- 
tion of persons with similar values is rewarding, resulting 
in effective communication and a minimun of tension; and (3) 
rewards leave each person with a feeling of satisfaction 
with his or her partner and thus a desire to continue the 
relationship (Coombs, 1961) . It is Coombs' (1966) belief 
that because of this emotional aspect it seems reasonable to 
expect that persons will seek their informal social rela- 
tions with those who uncritically accept their basic values 
and thus provide emotional security. Such compatible compan- 
ions are most likely to be those who feel the same way 
about important things, i.e., those who possess similar 
values. 

It is Coombs' (1966) contention that the theory of 
value consensus helps explain the findings that exist deal- 
ing with homogamy, endogamy, propinquity, complementary 
needs, or ideal-mate conception. Sharing values brings 
people together, both spatially and psychologically. Thus, 
a person may want to marry a member of the same racial 



- 10 - 



group, for example, because this might be a very impor- 
tant value in and of itself, or because persons who share 
similar social backgrounds will likely be socialized under 
similar conditions and conseqently develop similar value 
systems . 



Stimulus Value Role Theory 

The stimulus-value-role theory suggests mate selection 
is not opposites attracting but an exchange process that 
evolves through stages. Murstein (1970) suggests that in a 
free-choice (or open-field) mate selection situation, such 
as in the United States, the selection of a marital partner 
occurs in a three-stage process. In the stimulus stage, one 
person is drawn to another because of his or her perception 
of the other's physical and social attributes. If the 
couple is fairly matched, the relationship may develop on 
into the value stage. The value comparison stage occurs 
when the couple begins to assess their compatibility by 
exploring and testing each other's attitudes toward life, 
sex roles, and other salient belief areas. This stage in- 
volves the appraisal of value compatibility through verbal 
interaction. If, as they discuss these and other areas, 
they find that they have very similar value orientations, 
their feelings for one another are likely to develop. Some 
couples marry at this point, but most move on into the role 
stage. 



- 11 - 



The role stage is comprised of the expectations of the 
individual, his or her partner, and the perceived fulfill- 
ment of these expectations. Now the couple's association is 
so continuous that they not only hear one another's expres- 
sed values, but they see how those values are expressed in 
real life situations. Though some individuals marry only on 
the basis of the stimulus and value compatibilities, the 
majority of persons move into the role stage. Murstein 
(1976) suggests that successful passage through the three 
stages generally leads to some degree of permanence in the 
relationship. Additional research indicated support for an 
equity position and for role compatibility over similarity 
in fostering courtship. 

Criticisms of the theory have been directed at the 
proposed sequencing of variables. Additionally, some re- 
searchers argue that the value stage has not been success- 
fully demonstrated and will be necessary to provide suffici- 
ent evidence of its existence (Rubin and Levinger, 1974) . 

Filter Theory 

A fourth approach incorporates the role of social 
structure in mate selection. Kerchoff and Davis (1972) 
tested both the value consensus and the complementary needs 
theories with the assumption that they might operate at 
different stages in the courtship process. They concluded 
that there are various "filtering" factors operating 



- 12 - 



sequentially throughout the mate selection process and 
acting to reduce the number of potential partners. 

Initially, social attributes such as religion, educa- 
tion, and the father's occupation are in operation to filter 
out ineligible spouses. After the operation of social demo- 
graphic factors has established a pool of eligibles, the 
final selection of a partner is then based on more indi- 
vidual characteristics such as personality. As the relation- 
ship continues, a consensus of values becomes significant 
and complementary needs play an important part. Kerchoff and 
Davis (1972) suggest that the complementarity factor is only 
seen later in the relationship because of unrealistic 
idealization of the loved one in the earlier stages of the 
relationship. Mate selection is clearly a process deter- 
mined by a complex assortment of social factors that both 
restrict and enhance choice. 

Each of the mate selection theories discussed above 
leaves unexplored the role which age plays in mate selection 
and focuses on the psychological factors that contribute to 
attraction and marriage between individuals. A discussion of 
the psychological dimensions expresses the individual's per- 
sonal level of involvement. By this focus, the theories 
overlook a number of social structural forces that affect 
the selection of a mate. Typically, the individuals in- 
volved are not aware that these forces are creating or eli- 
minating a selection market for them. Consequently, they, 
as well as social scientists, have primarily centered on 



-13- 



psychological rather than sociological processes of mate 
selection. Thus, the theories do not fully address the soc- 
ial structural constraints that are placed on the individual 
nor do they consider the variables the individual feels most 
important in their selection. Consequently, the continued 
effect of these variables has been discounted or viewed as 
less relevant. To the extent that these effects have been 
measured, it has been through the effects of individual 
individual level variables. In fact, it is felt that the 
influence of assortativeness for age, education, etc. has 
been well-documented and that other social structural con- 
straints such as race and ethnicity have declined in assor- 
tativeness (Murstein, 1980) . This research contends that 
other scientific studies have neglected to assess the conti- 
nued significance of the structural variables in mate selec- 
tion. Consequently, a number of the studies on age discre- 
pant marriages have been based on the psychological rather 
than the sociological perspective (Cuber, 1971; LaPatra, 
1980; Houston, 1987). 

This analysis will attempt to assess the effects of the 
structural variables of age, race, education and ethnicity 
on the likelihood of black and white women's participating 
in age discrepant marriages. Throughout the literature 
there is evidence of a widespread tendency for individuals 
to marry others who are like they are; therefore, these 
social structural variables will be analyzed on the basis of 
similarity or dissimilarity. In this research, 



these 



-14- 



variables are referred to as "homogamous" and "heterogamous" 
(Burgess and Wallin, 1949; Rockwell, 1976; Dressel, 1980). 



CHAPTER THREE 

HOMOGAMY AND HETEROGAMY IN MATE SELECTION 

Homooamv 

The homogamy hypothesis suggests that people are most 
attracted to others who share similar characteristics such 
as age, race, religion and social class. A basic theory of 
homogamy is Fritz Heider's (1958) balance theory. Generally, 
the balance theory maintains that less stress is produced 
when one selects a partner who has the same beliefs, 
thoughts, and feelings about other persons, ideas, events 
and similar characteristics such as race, religion, and 
social class. 

One of the basic determinants of homogamy is geographic 
proximity, which means quite simply that individuals are 
most likely to meet and to date individuals who live or work 
nearby. Since residential patterns in the United States 
tend to be characterized by racial and socioeconomic segre- 
gation, those who live nearby tend to share similar charac- 
teristics. Another reason for homogamous mating is the in- 
tense social pressure, from family and friends, to marry 
from within "one's own kind." This pressure results from 
the fact that homogamy functions to maintain the status quo 
(e.g., to ensure that wealth and material possessions will 



- 15 - 



-16- 



remain within one's family and to preserve traditional 
values and beliefs) . Additionally, people generally feel 
more comfortable with others from similar backgrounds. They 
may have less to argue about, as well as more in common to 
talk about and enjoy together. Differences may make com- 
munication between individuals from different social groups 
difficult. 

The notion of complementarity seems to be contradictory 
to the theory of homogamy, probably because both have been 
overgeneralized. Most couples are likely to be homogamous 
in several socio-demographic characteristics, but complemen- 
tary in many idiosyncratic ones. They may be alike in edu- 
cation, religion, and family income backgrounds, for in- 
stance, but have complementary differences in certain areas 
of interpersonal intimate behavior. 

Given the strong societal pressures that operate to 
ensure that individuals marry from within their own social 
categories, coupled with the relatively lower probability of 
meeting prospective dating partners who are dissimilar, it 
is surprising that marital heterogamy occurs at all. Yet, 
recent Census statistics indicate that about 10 percent of 
all intact marriages are religiously heterogamous, that one 
percent are racially heterogamous, and that 10 percent are 
age heterogamous, in that the spouses' age differ by four 
or more years (U.S. Census, 1985). 



-17- 



Aae Homoaamv 

One of the major dimensions of homogamy in mate selec- 
tion is spouse's age. People similar in age have much in 
common (Riley and Waring, 1976) . They share many ideas and 
experiences because they are at similar points in the life 
course. They belong to the same cohort, and they are likely 
to play similar roles (Uhlenberg, 1975) . The age structure 
makes individuals who are situated within it develop cul- 
tural emphases, social behavior patterns, and a certain 
psychological bent (Merton, 1968) . With respect to mar- 
riage, the individual is led to believe that failure to 
select a spouse who shares one's historical context may be 
an invitation to marital instability or even failure in the 
marriage . 

In our society, although spouses are expected to be 
members of the same generation, husbands are expected to be 
a few years older than their wives . For example, in 1890 
the mean age difference between groom and bride in the 
United States was approximately four years for first marri- 
ages. Specif icallly, 26.1 was the mean age for the groom 
groom and 22.0 is the mean age for the bride. By 1950, the 
difference had dropped to about two years. The mean age for 
the groom was 22.7 and the mean age for the bride was 20.3. 
This same age differential has been found among all marri- 
ages as late as 1982 (26.7 and 24.4). In a midwest county, 
a study was designed to examine the ages of first time 



-18- 



married couples during a fifty year period. For the sample 
of 275 marriages examined, males were married for the first 
time at a mean age of 22.5 years, while females were married 
at a mean age of 20.2. In 75.3 percent of the marriages 
examined, the male was older than the female. In 15.6 per- 
cent of the marriages, both partners were the same age, and 
the female was older in only 9.1 percent of the marriages. 
Thus, males were significantly more likely to be older 
at the time of first marriage (Patterson and Pettijohn, 
1982) . 

The factors accounting for this difference are numer- 
ous, but the following represent major causes: (1) the tra- 
ditional responsibility of the husband to be the major 
breadwinner, which requires time for preparation and post- 
pones marriage somewhat, (2) the slight excess of males 
relative to females at all ages through the late 20' s and/or 
(3) the mating gradient ( the tendency for males to marry 
down and females to marry up in age, social class, and edu- 
cation) . In recent history the bride has been younger than 
the groom in about three quarters of all marriages, while 
brides are older in about 10-14 percent of marriages 
(Bytheway, 1981; Berardo, Vera, and Berardo, 1983; Goldman, 
Westoff, and Hammerslough, 1984; Vera, Berardo, and Berardo, 
1985; Gunter and Wheeler, 1986). In 1980, the male was 
older than his bride in 69 percent of all the married 
couples in the United States (National Center for Health 
Statistics, 1983) . 



-19- 



The magnitude of the age differential historically has 
varied with age at marriage. The discrepancy increases as 
male marriage age increases, starting at about one year's 
difference among males marrying at age 20 and increasing to 
as much as ten years among males marrying at age 60. Viewed 
from the perspective of the woman's age at marriage, the age 
differential declines from about five years at age 15 to 
less than one year at age 30. Beyond 30, the gap begins to 
widen again, reaching two to four years (male older) among 
women marrying at older ages (Leslie and Korman, 1985:370). 

Age homogeneous patterns reflect the socialization 
process and institutional norms that most Americans are 
exposed to throughout their lives. However, age homogamy in 
mate selection apparently becomes less salient over the life 
cycle. "As people grow older they become more and more dif- 
ferent from one another, so that age means less and less as 
a determinant of character and behavior" (Baltes, 1980:74). 
For example, most people would consider a 37-year-old man 
who marries a 32-year-old woman to be within the normative 
guidelines of marriage partners. However, if a college 
sophomore proposed marriage to a freshman girl in high 
school, one can safely predict social disapproval even 
though the age gap between the two students equaled that 
between the 37 and 32-year-old partners. Even greater dis- 
approval would occur if the college sophomore were a female 
and the freshman in high school were a male. 



- 20 - 



Aae Heteroqamv 

An empirical study of marital age heterogamy showed 
that in 1980, 27 percent of males were five or more years 
older than their spouses. In constrast, only 3.1 percent of 
the males were five or more years younger. In both types of 
age heterogamy (husband older as well as husband younger) , 
when compared with homogamous age marriages (operationally 
defined as plus or minus four years) , a tendency existed for 
the heterogamous marriages to be characterized by lower 
educational levels, multiple marriages, lower family status, 
and lower occupational status of the husband (Atkinson and 
Glass, 1985) . 

Cowan (1984) attempted to determine the effects of 
degree of age discrepancy and sex of the older person on at- 
tributions of success for relationships by varying the sex 
of the older person. He used definitions of moderate age 
discrepant relationships (seven years) and highly discre- 
pant age relationships (18 years) . Both college and high 
school students judged the potential compatibility of a 
fictitious couple. Evidence for the double standard was 
clearest when there was a large age discrepancy. The 
students felt that it was socially acceptable for the hus- 
band to be much older than the wife than to have the wife 
significantly older than the husband. However, it should 
also be noted that the high school males rated the large age 
discrepancy condition less positively regardless of who was 



- 21 - 



older. Furthermore, high school students responded more to 
the age discrepancy itself than to which sex was older, 
attributing less chance for success to couples with a large 
age gap (Cowan, 1984) . 

With respect to sex differences, both samples consis- 
tently indicated that males did not show the double standard 
when the discrepancy was only seven years, yet the males 
rated the relationship less positively than females when the 
discrepancy was 18 years. It was provocative that the 
female, who limited her options by acceptance of the double 
standard, held it as strongly at the level at which it was 
more probable to occur, at a moderate, level, as when the 
discrepancy was large. Thus, with only the exception of male 
responses to a moderate age discrepancy, the double standard 
manifests itself in less optimistic views toward relation- 
ships in which the female is older than the male (Cowan, 
1984) . 

Recent research on age difference in marriages presents 
three patterns of interest: (1) The percentage of men 
marrying older women shows two peaks at age 20 and again in 
the late twenties (28) ; the rise to the latter peak begins 
at about age 23. (2) Women display a gradual, through 
persistent, rise in the proportion marrying younger men, 
reaching a plateau in the late thirties and early forties; 
the major change for women appears to be in the late 
twenties (at age 28) and continues relatively high to peak 
at ages 41 and 48, with over 61 and 60 percent, 



- 22 - 



respect ively , marrying younger men after these ages. The 
percentage appears to decline slightly thereafter. (3) Men 
more often exhibited traditional age discrepant marriages 
after the ages of about 31 or 32 (Gunter and Wheeler, 1986) . 

Thus, a gradual shift toward greater toleration of 
older women marrying younger men may eventually emerge 
through a variety of factors such as (1) the women's move- 
ment, (2) increased female participation in the labor force, 
(3) greater economic parity for women, (4) higher educa- 
tional attainment by women, and (5) the sex ratio. The 
women's movement has affected both sexes by providing new 
role expectations and responsibilities, including a greater 
role for women in selecting mates — countering the idea that 
they are passive participants in courtship. Some textbooks 
and popular magazines, e.g., Cosmopolitan and Essence's, now 
include discussions about what women look for in a mate and 
not just what a man would consider an ideal mate. In 1980, 
more women were in the civilian labor force than those who 
remained home. The labor force participation rate for women 
rose from 37.7 percent in 1960 to 43.3 percent in 1970. By 
1980, the rate had risen to 51.6 percent, passing the 50 
percent mark set in 1978 (United States Bureau of the Cen- 
sus, 1981) . 

As women began to enter employment in greater numbers 
and to seek positions that were previously held by men, a 
corollary increase occurred in their social, business and 
political participation in the society. This broadened the 



-23- 



basis of mate selection for women. The increase in employ- 
ment opportunities and emerging career opportunities indi- 
cates for women expanding long-term options outside of the 
family. Such options make it easier for them to choose to 
remain single longer and marry at a later age. Many of 
their marriages at this point may fall into the age dissimi- 
lar category because they are marrying later than male mar- 
riage cohorts. 

Participation in the economic marketplace also in- 
creased women's financial resources. While parity has not 
been has not been totally achieved, greater economic 
resouces among women have changed the nature of heterosexual 
relations, including the mate selection process. Today, 
women are less apt to marry for economic security because 
they can achieve independence through their jobs or profes- 
sional careers. Because they tend not to marry while in 
college, and because education better eguips them to compete 
in the marketplace, more and more women have delayed mar- 
riage (Newsweek, 1986) . 

Another reason why women may be inclined to participate 
in age dissimilar marriages is influenced by the sex ratio 
factor. The significance of the sex ratio is this: If 
there is a shortage of members of one sex at the usual ages 
of marriage, members of the other sex may be forced to go 
beyond the age range dictated by the homogamy norms in order 
for them to find an acceptable partner. However, Mensch 
(1986) researched the trends in age difference for those 



-24- 



born between the early 1930's and 1950's and examined fac- 
tors which might account both for the apparent changes that 
have occurred during the period when these birth cohorts 
were at risk of marriage and for the differentials that 
exist at any one point in time. The results of her research 
indicated that the age of the wife at marriage is inversely 
related to a couple's age difference. That this might sim- 
ply be due to the age distribution of available men was 
considered and rejected. It was speculated that the rela- 
tion between age difference and age at marriage is a conse- 
quence of changing preference, not of the supposed shortage 
of suitable single men. 

Results from questionnaires on age and the choice of a 
mate distributed by Pietropinto and Simenaurer (1977) to 
four thousand men indicated the following: 



Age does not matter at all 38.2% 

The woman must be my age or 

younger 16.4% 

The woman could be 5 years or 

older, not more 16.3% 

Age doesn't matter, but she 

must not look older 12.8% 

The woman could be 10 years 

older, not more 11.3% 

Prefer an older woman 3.0% 

No answer 2.0% 



Note that at least forty percent of the men would 
accept an older woman as their mate without reservations. 
Only 16.4 percent indicate the mate must be the same age or 
or younger. Clearly, men in this study did not see age 
difference as a major deterrent to their selection of a 
spouse. Even when the respondents were asked if women could 



-25- 



be five years older, but not more than 10 years, the rates 
for responses decreased but not drastically. 

There are two distinct ways in which these norms of age 
homogamy in mate selection can be violated. The first in- 
volves women's marrying younger men. The second involves 
women's marrying men who are considerably older. These 
types of age heterogamy are caused by different factors and 
are regarded differently by our society. A discussion of 
each type follows. 

Women Who Marry Older Men 

It seems more acceptable in our society for younger 
women to marry considerably older men than it is for women 
to marry younger men (Sontag, 1972) . As men age, they do 
not inevitably lose their attractiveness and desirability, 
but as women grow older, they generally are considered less 
attractive and less desirable. Attractive men can remain 
eligible well into old age. They are considered acceptable 
mates for women their own age as well as for younger women. 
Conversely, for women, even if they are attractive, "aging 
means a humiliating process of gradual sexual disqualifica- 
tion" (Sontag, 1972) . Beauty, "identified, as it is for 
women with youthfulness, does not stand up well to age" 
(Sontag, 1972). As a result, women in our society become 
sexually ineligible much earlier than men do. Consequently, 
Laws and Schwartz (1977:70) indicated that 



-26- 



the older woman is expected to be 
permanently out of circulation . . . 
she can be a respectable married lady 
and a mother or widow, or even a grand- 
mother, but she is not permitted to be 
a sexual person, because no man admires 
her. 

This double standard of aging operates as a mating 
gradient, which functions to the advantage of older men and 
the disadvantage of older women in the delineation of a 
normative field of eligibles. The mating gradient is a term 
generally used to refer to the tendency for men to marry 
women of lower status and for women to marry men of higher 
status (Leslie and Korman, 1985) . As a function of the 
mating gradient, upper class women will receive approval 
from their parents and peers only if they marry someone of 
equal or higher status. On the other hand, approval is 
usually not withheld from males who marry below themselves 
in age and status. Such pairing results in some high status 
women and low status men remaining single. When applied to 
age, the mating gradient means that some older women might 
have a restricted pool of eligibles. 

In discussing why women marry men ten or more years 
older, Seidenberg (1972:9) argues women can usually obtain 
economic and social security with many years of struggle; 
however, it would take men fewer years to accomplish the 
same level. Women can rarely achieve status at a rate 
which men usually can achieve on their own. It remains to 
be tested whether, as Seidenberg (1972) assumes, this pat- 
tern of age discrepancy typically involves wealthy older 



-27- 



men. But implicit in this thesis is the derived nature of a 
woman's status: to achieve high status, a woman must marry 
a high status man. 

Popular magazines often contain articles about younger 
women marrying older men. Invariably, these women are less 
affluent and are marrying men who are exceedingly affluent, 
have high status occupations, or, in a small number of 
cases, severely violate the age norms to the point where 
these marriages are illegal. Illustrations of such mar- 
riages include a 23-year-old bride marrying U. S. Supreme 
Court Justice William 0. Douglas, who was in his 68th year? 
a woman 39 years old marrying Herbert A. Armstrong (founder 
of the World-Wide Church of God) , who was 84 years old; a 
35-year-old woman marrying Edward Steichen, the world re- 
nowned photographer, who was 88 years old at the time of 
their marriage; a 45-year-old woman marrying General Omar 
Bradley, who was 74 at the time of their marriage (West, and 
Hotchner, 1967) ; and a 16 year-old student marrying Ed 
Christoph, her 37 year-old English teacher. Because of his 
relationship with her and their subsequent marriage, he was 
fired from his teaching position (Gainesville Sun, 1983) . 

However, some studies suggest that in marriage, younger 
women provide greater benefits for older men (Rose and Bell, 
1971; Fox, Bulusu, and Kinlen, 1979). Both research groups 
found that marriage to a younger wife was significantly 
related to a husband's longevity. In fact, Foster, Varta- 
bedian, and Wispe (1984) completed an additional study and 



-28- 



found that analysis of mortality estimates and adjusted 
census data showed that men aged 70 to 79 married to younger 
women tended to live longer than men married to older women. 
Two possible explanations were advanced for this relation- 
ship: (a) premarital selection factors may be related to 

longevity and (b) psychological, physiological and social 
factors associated with longevity may be enhanced by mar- 
riage to younger women. 

Women Who Marrv Younger Men 

The pattern of older women/younger men is worthy of 
observation because it has been a "taboo" in our society. 
Newspapers periodically carry examples of marriages that 
have greatly exceeded a pattern. They are typically con- 
sidered by the publishers as newsworthy and perhaps of some 
social concern to the community. An example appeared in the 
"Newsmakers" column of the Tampa Tribune (1982: 28-A) : 

Fulvio Cerutti, 19 married 85 year-old Maria 
Pia Cerutti Thursday in a Omegna, Italy, City 
Hall ceremony cheered by hundreds of towns- 
folk. 

The white-haired bride and the handsome 
Cerutti, a factory storekeeper related to the 
local nobility, refuted suggestions that 
financial interests were involved by specifi- 
cally stating at the civil ceremony that they 
did not want to share their possessions. 

The bride owns considerable real estate 
in Omegna, a prosperous town near Lake Maggiore 
in nothern Italy. 

It is obvious that couples with age differences that 
are 50 or more years represent relationships that are truly 



-29- 



discrepant. The above couple clearly deviates from the 
standard pattern. The marriage partners also exceed the 
range where they could share what Riley considers as the 
historical context of life (Riley and Waring, 1976) . Of the 
2 million weddings performed each year in the United States, 
22 percent are between older women and younger men, up from 
16 percent in 1970 (Toufexis, 1987) . In Hillsborough 
County, Florida between June 1,1984 and March 30, 1985, ages 
for both marriage partners were recorded from their marriage 
license applications. In almost one-quarter (24.7%) of the 
marriages, the bride was older than the groom (Gunter and 
Wheeler, 1986) . This is almost twice as high as the propor- 
tion reported in earlier literature (Glick and Landau, 1950; 
Berardo, Vera and Berardo, 1983). Even when they expanded 
the definition of age homogeneity to include all marriages 
in which the partners are within three years in age as 
suggested by others (see Vera, Berardo and Berardo, 1985) , 
they found a substantial proportion of the marriages (14%) 
fell into the female older category. 

The following characteristics have been observed among 
the older women/younger men couples: (1) the woman is usu- 
ally over 30, divorced (from an older man) and often has 
children (Houston, 1987) , and (2) her younger partner typi- 
cally was reared by a working mother and has sisters who 
also have careers. In addition, the following patterns were 
revealed as occurring (Houston, 1987) : 



-30- 



Women in their twenties are attracted to, 
living with, or marrying men who are one 
to eight years younger. 

Women in their thirties and forties are drawn 
to men as many as 10 to 18 years younger. 

Women in their fifties and sixties show the 
greatest age differences from their boyfreinds 
or spouses, as there the span between them 
could be as much as 25 years. (p.ll) 

Although the popular media gives the impression that 
relationships between older women and younger men represent 
a new trend, this may not be the case (Atkinson, 1982) . In 
an analysis of two national data sets, using the relative 
distributions of husband and wives in 1900, 1960, and 1980 
and total number of marriages contracted in 1950 compared to 
1977, Atkinson (1982) found that 27 percent of the husbands 
were younger in 1900, 15 percent in 1960, and 30 percent in 
1980. She contends that this change is an artifact of dif- 
ferential mortality rates and sex ratios rather than a trend 
in mate selection. When the total number of marriages con- 
tracted in 1950 is compared to those contracted in 1977, it 
appears that there has been only a slight increase in the 
number of older women/younger men relationships. 

Seskin and Ziegler (1979) attempted to explore the 
relationships of older women and younger men. One of their 
most interesting findings was that men who sought out and 
courted older women had had a very pleasurable sexual exper- 
ience with a much older woman when they were adolescents. 
In fact, in most cases, it was their first sexual inter- 
course. These men appeared to need more mature women than 



-31- 



they could find in their own age groups. Such relationships 
allowed the men to complete the process of growing up, in 
some cases. In others, the women provided the men with a 
non-threatening, mature relationship wherein they did not 
have to assume the role of the traditional husband with the 
responsibilities of parenting and child-rearing. 

It has also been suggested that men who seek out older 
women do so because they are frightened by the increase in 
sexual freedom and the aggressiveness of younger women. 
However, the same risks appear to be present for all indi- 
viduals who enter into interpersonal relationships. Matur- 
ity appears to be a major underlying factor for success or 
failure in intimate relationships (Seskin and Ziegler, 
1979) . 

Derenski and Landsburg (1980) studied 50 couples in- 
volved in older women/younger men relationships. They 
sought to shed light on the socio-psychological factors that 
might affect these relationships. The questions of "going 
public," "breaking rules," "power in the relationship," and 
"who's looking for a mother" were all explored through the 
personal statements of the respondents. The authors con- 
cluded that the qualities that attract older women and 
younger men to one another are financial and emotional de- 
pendence, role flexibility, ability to discard inappropriate 
rules, and willingness to share power. 

Since their age pattern is contrary to popular prac- 
tice, couples in an older women/younger men marriage 



-32- 



apparently value their personal relationships more than 
conformity to custom. In other words, their motivation to 
be married to each other may be greater than (or, at least 
different from) that of those who follow the usual homog- 
amous pattern. Another important consideration for women 
entering a relationship with a younger man is sexual inti- 
macy. Other factors include feelings that the man in an 
older women/younger men relationship is more considerate 
than in other relationship previously encountered, provides 
more excitement and shows more acceptance of the women for 
themselves, and does not treat them as objects (Derenski and 
Landsburg, 1980) . 

It has been found that women who marry younger men 
also stay married longer (on the average) than women who 
marry older men. On the average, an additional year could 
be added to the length of the marriage for each year the 
wife is older than her husband (Burgess, Locke and Thomas, 
(1971) . This is probably less a magic formula than an in- 
sight into the personalities of the participants. Accord- 
ingly, Burgess, Locke and Thomas (1971) found that marriages 
are less likely to end in divorce if the bride is older 
than the husband than if the reverse is true. They suggest, 
therefore, that if the goal of permanent marriage is an 
important value, a man is most likely to achieve it by mar- 
rying a woman older than he some time after his twenty- 
eighth birthday. 



- 33 - 



Summarv 

The literature has clearly indicated that social pres- 
sures developed through the culture encourage a person to 
marry someone from his or her own social group. Sociodemo- 
graphic characteristics such as education, occupation, race, 
and age serve as the initial filters in the mate selection 
process and act to channel individuals into homogamous mar- 
riages. 

Younger women/older men unions have previously been 
considered financial and youth exchanges. Recent work, 
however, indicates that these relationships occur frequently 
in the lower income groups and are not concentrated only in 
the higher income strata. Contrary to popular notions, older 
women/younger men relationships are not necessarily a new 
trend. Perhaps because the proportion of the age discrepancy 
diminishes, differences become less salient over time. There- 
fore, older individuals may participate in age discrepant 
marriage without negative social sanctions. 



CHAPTER FOUR 

SOCIOCULTURAL DETERMINANTS OF AGE 
DISCREPANCY IN MARRIAGES 

Some of the strongest forces influencing marital 
choice in our culture are social demographic factors that 
encourage a person to marry someone from his or her own 
social group. These social demographic factors are age, 
race, religion, education, economic and ethnic background. 
This research will focus on race, education, ethnic back- 
ground and marital history factors as determinants of the 
likelihood of women's participating in age discrepant 
marriages . 



Racial and Ethnic Differences 

The choices of whether to marry and at which age to 
marry are not only influenced by economic, racial and social 
trends, but the extent of the impact depends crucially on 
cultural differences. Anderson, Hill and Butler (1987) 
studied the Malaysian pattern of mate selection. They used 
a proportional hazards model to estimate the age at which 
women would marry. Their results indicated that the age at 
marriage of Malaysian women responded significantly to 
economic variables. Therefore, skilled employment of both 



- 34 - 



-35- 



husbands and wives, expecially professional employment, 
delayed marriage relative to husbands and wives who worked 
in unskilled employment and nonemployment. However, race was 
very important in explaining differences in marriage pat- 
terns in Malaysia. Chinese women tended to marry later 
than native Malay or Indian. 

Racial 

Spanier and Glick (1980) noted that in American several 
demographic patterns restrict American black women's pool of 
marriage eligibles more than white women's and that black 
women are more likely than whites to vary from the 
traditional pattern of mate selection with respect to age 
homogamy. This was most pronounced in cases where the 
husband was at least four years older than the wife. 

Even at ages 20-24, American black women seem to be at 
a disadvantage, with an availability ratio that shows 93 men 
per 100 women. The difference between whites and blacks at 
this age may reflect differences in the United States Census 
undercounts that have not been adequately adjusted for 
single black males in their twenties and thirties. Black 
women appear to be at a greater disadvantage than white 
women until age 45. The probable inaccuracy of the Census 
counts for unmarried black males makes one regard the esti- 
mates for blacks as biased in the direction of underesti- 
mating prospects for younger black women (Goldman, Westoff, 
and Hammerslough, 1984) . 



-36- 



Black women outnumber black men in all age groups 
except the five to 14 group (Jackson, 1978) . For several 
reasons, this will become even more pronounced in the 
future: the majority of black births are female, the death 
rate is higher among young black males than females, and 
black men are entering the military in increasing numbers 
(Staples, 1978) . The shortage of young black men causes an 
especially large proportion of black women to marry men who 
are significantly older than the typical age at marriage for 
the population as a whole or to remain single (Spanier and 
Glick, 1980) . 

Ethnic 

Kearl and Murgia (1986) examine age differences among 
Hispanics and Anglos spouses, both homogamous and heterog- 
amous with regard to ethnicity. When looking at the fe- 
male's age at marriage they find the age differences to be 
greatest when Spanish Surname women marry non-Hispanic men. 
It is approximatley 6.3 years. The discrepancy is least 
when both spouses are Hispanic. The difference is only 
about one year when the females age is in the 24 to 26 
range. When looking at the males 's age the difference 
tends to increase with age. Above about 33, the discrepancy 
is about nine years for couples either or both spouses are 
Hispanic. 

Murguia (1982) has observed that Spanish surnamed fe- 
males with a divorce history were more likely to marry 



-37- 



exogamously, while previously divorced Anglo females were 
less likely to marry exogamously. 

Education/Social Class 

The marriages of couples from different age cohorts 
are not distributed randomly in the population; they are 
most frequently found in what could loosely be called the 
upper-middle class (Cuber, 1971 and LaPatra, 1980) . A pos- 
sible explanation could be that these people have arrived at 
a stage in their lives where they are no longer striving for 
additional status and feel that a marriage of this type 
would not negatively affect them. However, Vera, Berardo, 
and Berardo (1985) found that the notion that age dissimilar 
mates are most prevalent among the middle and upper classes 
was not supported by their research. In fact, according to 
their analysis, age discrepant unions were clearly more 
prevalent among the lower class grouping, as measured by 
income and socioeconomic status. 

One indicator of social class is education. Existing 
research suggests how education is related to women's like- 
lihood of entering an age discrepant marriage. As women 
increase their levels of education, they are delaying the 
time at which they will get married. Consequently, women 
are getting married later. This situation increases the 
possibility of their becoming candidates for age discrepant 
unions . 



-38- 



In the past, women usually married men with the same or 
higher level of education. One reason for this was that 
until recently, significantly more men than women had com- 
pleted college. In addition, because women generally mar- 
ried men who were the same age, or older, their spouses were 
likely to have achieved higher levels of education. Rockwell 
(1976) used the 1970 U. S. Census to determine trends and 
variations in educational homogamy and husband-hypogamy . He 
found that educationally homogamous marriages were occurring 
at levels higher than expected by chance but lower than 
would occur if equality of education were determined in mate 
selection. In addition, he suggested that a trend for white 
women in racially heterogamous marriages to marry up in 
education is also related to change in the relative educa- 
tion of men and women. 

Now that more women than men are graduating from high 
school and women are just as likely as men to finish col- 
lege, it is possible that the earlier pattern of education 
heterogamy will change. Moreover, as more women opt to have 
careers which require increased years of education, they are 
placing themselves in a different marriage market, insofar 
as they will be limited to a smaller pool of better educated 
men. Doudna (1981:22) indicates, 

These contradictory patterns create a 
crunch which has been exaggerated by the 
rapid emergence of the new class of 
single professional women. There has been 
a sudden growth on the demand side of the 
market and very little movement in the 
supply lines. The elite have become 
demographic losers; they've priced 



-39- 



themselves out of the market. The problem 
that used to concern only the heiresses — 
where to find a suitable mate among 
the sparsely stocked and heavily fished pool 
of men at the top — affects an entire class. 

Because college-educated women tend to restrict them- 
selves to partners of equal or greater educational attain- 
ment, their marriage market is smaller than that for women 
whose education ended after high school. Compared with 
women who discontinued their education after high school, 
college-educated women suffer an additional market disad- 
vantage of around 25 percent (Westoff and Goldman, 1984) . 
For this reason, then, one can expect that more highly 
educated women have a greater likelihood of marrying outside 
the conventional age categories. 

Constraints on the educational level of potential 
spouses alter the available pool of mates, particularly for 
highly educated women and for black women (who on the 
average, have more years of education than black men) . The 
marriage market for college-educated black women is par- 
ticularly limited, with measures of availability in major 
urban areas reaching values as low as 1:5 men per woman 
(Goldman, 1977) . 



Marital History 

Previous marital status has also proven to be signifi- 
cant in explaining age differences between spouses. Glick 
and Landau reported in 1950 that with first time marriages 
for both spouses the age difference was 2.8 years. If the 



-40- 



husband or wife were remarrying the average difference in- 
creased to 4.7. In marriages in which neither spouse had 
ever been previously married, the bride was usually one to 
three years younger than the groom. If both spouses had 
been previously married, their age difference was, on the 
average, four years. If a bride has never been married but 
her husband has, he will be, on the average, six years older 
than she. If a man has never been married but his bride 
has, the average difference between their ages is only a few 
months (United States Department of Human Services, 1987) . 

Contrary to popular myth, very few divorced men marry 
women who are a great deal younger than they are. Only one 
divorced male in 25 married a woman 20 or more years younger 
than he. It is even more rare for a divorced woman to marry 
a man slightly younger than she (Peters, 1976) . Contrast- 
ing 1980 with 1983 the data indicate the median age of the 
grooms decreased by .2, but it was still greater than that 
of the brides. When single brides marry divorced men, the 
difference between median ages only increased by .1. When 
both partners had been divorced, the median age difference 
decreased by .4. More significantly noted, however, was the 
difference in the median age of divorced brides marrying 
single men. The median age difference increased from .9 to 
1.2 years (United States Department of Human Services, 
1987). There are at least three reasons why remarriages are 
more likely to be age discrepant. First, people who remarry 
are older than first timers, and they feel less compelled to 



-41- 



have age peer mates. Second, a poor previous marital ex- 
perience may lead a person to seek an age dissimilar part- 
ner. Third, as people move through the life cycle, they 
tend, generally, to feel less constrained by rules, includ- 
ing those impinging of the selection of a mate (Berardo, 
Vera and Berardo, 1983) . Thus, it appears that a previous 
marriage of one or both spouses increases the likelihood of 
an age heterogamous union. A contributing factor here is 
that the previously married (at least in the case of men) 
have a wider range of younger people from whom to choose. 

Sex Ratio 

As previously noted, an unbalanced sex ratio may result 
in more people marrying outside appropriate age categories. 
Historically, the United States has had more men than women. 
This was due to a differential migration of men and women to 
this country and, to a lesser extent, to the fact that a 
considerable number of young women died in childbirth. 
Today, this situation is reversed due to changes in immigra- 
tion patterns and greater improvement in women's than in 
men's health. Since World War II there have been more women 
than men. In 1970, for example, there were only 95 men for 
every 100 women; in 1910, there were nearly 106 men for 
every 100 women. The data indicate that beginning at age 
35, there are fewer unmarried men than women. If each of 
these people over age 35 wanted to marry, many women would 
be left; this is even more true in the older age groups. 



-42- 



The disparity between the number of men and women in the 
marriage market increases greatly with age. A related 
picture of marriages is provided when one compares ages of 
grooms with ages of brides, i. e., age differences at mar- 
riage and different ages of grooms. The proportion of 
grooms marrying older women remains fairly constant between 
15 and 18 percent at each age level from 20 to 64. Also, 
the proportion of grooms marrying brides in the same age 
group has remained at the same level. But the proportion 
of grooms marrying women eight or more years younger in- 
creased from 24 to 46 and 48 percent from ages 35 to 75 
and over (Department of Health, Education and Welfare, 
1977) . 

From a demographic perspective, the disparities in the 
relative number of eligible men and women can be great, 
depending on the metropolitan area one lives in. From the 
moment of birth, death rates of males exceed those of fe- 
males. This is largely because of the relatively greater 
numbers of deaths among men from accidents and from cardio- 
vascular and respiratory diseases. Consequently, there are 
approximately 7 percent more female 50-year-olds than males 
and about 32 percent more female 70-year-olds (Westoff and 
Goldman, 1984) . 

From the research conducted in this area, three impor- 
tant issues emerge. First, what are the characteristics of 
the women who will most likely participate in age discrepant 



-43- 



marriages? Subsequently, it is important to ask whether the 
effects of these characteristics vary according to race. 
The final issue involves the notion that age discrepant 
marriages are likely to be heterogamous with regard to other 
social characteristics. 

As a result of these issues, the following hypotheses 
emerge and will be tested: 

HI: Black women are more likely than white women to 

be in age discrepant marriages. 

H2: Women with higher levels of education are more 

likely to be in age discrepant marriages. 

H3: Women in their first marriage are less likely 

to be in age discrepant marriages. 

H4: Women with Spanish ethnic backgrounds are more 

likely to be in age discrepant marriages. 

H5: Women in racially mixed marriages are more 

likely to be in age discrepant marriages. 

H6: Women in educationally heterogamous marriages 

are more likely to be in age discrepant 
marriages . 

H7 : Women in ethnically heterogamous marriages are 

more likely to be in age discrepant marriages. 

H8: Among black women the strongest predictor of 

participation in age discrepant marriages will 
be increased education levels. 

H9: Among white women the strongest predictor of 

participation in age discrepant marriages will 
be remarriage. 

Hypotheses will be tested for participation of women in 
any type of age discrepant marriage, for marriages in which 
the wife is at least five or more years older than her 
husband, for and marriages in which the wife is 10 or more 
years younger. This approach will test the effects of the 



- 44 - 



social and demographic variables for all women and then 
determine their effects for black and white women 



separately. 



CHAPTER FIVE 
METHODOLOGY 



Operational Defintion of Age Discrepant Marriage 

How much of an age difference must exist before a mar- 
riage is defined as age heterogamous or age discrepant? 
Researchers do not agree on this basic issue (Vera, Berardo 



and Berardo, 


1985) . Some 


consider any 


couple 


in which 


there is an 


age difference 


of greater than one year to 


be 


heterogamous 


(Jorgensen and 


Klein, 


1979) , 


while 


others 


re- 


quire larger 


differences , 


e.g. , 


three 


years 


( Jaco 


and 



Shepard, 1975; Gunter and Wheeler, 1986) and six years 
(Derenski and Landsburg, 1981) . Several researchers have 
followed a definition of age discrepancy which holds that 
the wife is ten or more years younger than her spouse, or is 
five or more years older than her husband (Veevers, 1984; 
Monahan, 1953; Glick and Landau, 1950). The present study 
will adopt this as an operational definition of age discrep- 
ant marriages because it represents a difference in which 
the age is at least 3 times the median age difference. Con- 
sequently, these couples have far exceeded the norms of 
"constrained age dissimilarity" typically found in marriage 
patterns. 



- 45 - 



-46- 



Data 

The data source was a one in a thousand public use 
sample of the person records from the 1980 United States 
Census of Population. From this sample of the population, a 
subsample of 46,844 married females was analyzed for age 
discrepancy. The subsample was selected by choosing house- 
holds of married women which included only the householder 
and the respective spouse. All other couples in the house- 
hold were deleted from the sample, i.e., adult children and 
their spouses who resided in the household. This elimina- 
tion was necessitated by the lack of complete information on 
relevant variables for other couples in the household. 

Although the research will include all married women, 
a major focal point will be the comparison of black and 
white women and their likely of participation in age 
discrepant marriages. Since the research will analyzed the 
independent impact of race and Spanish origin on the like- 
lihood of being in an age discrepant marriage, couples in 
which the householder identified Spanish as his/her race 
instead of an ethnic origin were excluded from the sample. 
Additionally, women who were 16 years of age or younger and 
heads of households were dropped when this procedure was 
used because the definition of heterogamy could not be 
suitably applied. The small proportion of those eliminated 
(100 out of 46,944) will not introduce bias into the 
analysis. The data collected for the householder and the 



- 47 - 



spouse were age, race, ethnic origin level of educational 
attainment, income, and times married. 



Variables 



Dependent Variables 

Acronyms and descriptions of the variables used in the 
analysis appear in Table 1. Initially, the major dependent 
variable is a dichotomy indicating whether or nor a woman is 
involved in an age discrepant marriage of either type 
(l=yes, 0=no) . This variable name is AGEDIS . 

The second measure of the dependent variable is a 
dichotomy indicating whether the woman is involved in an age 
discrepant marriage in which she is five or more years older 
than her spouse (l=yes, 0=no) . This variable name is 
AGEDISWO . 

The third measure of the dependent variable is a dicho- 
tomy indicating whether the woman is involved in an age 
discrepant marriage in which she is 10 or more years younger 
than her spouse (l=yes, 0=no) . This variable name is as 
AGEDISWY. 



Independent Variables 

The independent variables of race, ethnic origin, edu- 
cation, and first marriage were analyzed for their effect on 
a woman's likelihood of entering an age discrepant marriage. 



-48- 



TABLE 1 



DESCRIPTION 


OF VARIABLES IN THE ANALYSIS 


Variable 


Description 


AGEDIS 


RESPONDENT IS IN AN AGE DISCREPANT 
MARRIAGE OF EITHER TYPE, 1=YES 
0=NO) 


AGEDISWO 


RESPONDENT IS IN AN AGE DISCREPANT 
MARRIAGE IN WHICH THE WIFE IS 
OLDER BY 5 OR MORE YEARS 


AGEDISWY 


RESPONDENT IS IN AN AGE DISCREPANT 
MARRIAGE IN WHICH THE WIFE IS 
YOUNGER BY 10 OR MORE YEARS 


RACE 


1=BLACK 
0= WHITE 


GRADE 


22 GRADE LEVELS (0 TO COLLEGE 
COMPLETION) 


SPANISH 


RESPONDENT IS OF SPANISH ORIGIN 
CUBA, PUERTO RICO, MEXICO, AND 
OTHER SPANISH ( 1=YES , 0=NO) 


MIXRACE 


WIFE AND HUSBAND ARE OF 

DIFFERENT RACE (1=YES, 0=NO) 


MIXEDUC 


WIFE AND HUSBAND DO NOT HAVE THE 
SAME EDUCATIONAL LEVEL ( 1=YES , 
0=NO) 


MIXSPAN 


WIFE OR HUSBAND HAS A SPANISH 
ETHNIC ORIGIN 


FIRSMAR 


1=YES 

0=NO 


INCOME 8 


WIFES ' INCOME, IN ACTUAL DOLLARS 


MINCOME8 


HUSBAND'S INCOME, IN ACTUAL DOLLARS 


FAMINCOM 


TOTAL FAMILY INCOME, IN ACTUAL 
DOLLARS 


WORK 79 


WIFE WAS EMPLOYED IN 1979 (1=YES 
0=NO 



-49- 



Race is a dicohotomous variable, with blacks coded as 
(1) and whites coded as (0) . To determine whether racial 
heterogamy is related to age heterogamy, the variable 
MIXRACE was created. If the wife's race is the same as the 
husband's race, MIXRACE equals 0, and if their races are 
different, MIXRACE equals 1. 

Although the major dimension of comparison is between 
black and white couples, the impact of ethnic background is 
also of interest. Spanish origin was selected for special 
consideration because Hispanics represent one of the largest 
ethnic groups in the United States. In the 1980 Census of 
Population the question concerning Spanish origin included 
four countries of origin: Puerto Rico, Cuba, Mexico and 
Other Spanish speaking countries combined. In this analy- 
sis SPANISH is coded (l)if the woman is of Spanish orgin 
(0= not of Spanish origin) . 

Further analysis is also performed to ascertain the 
effect of dissimilar ethnic backgrounds, at least with re- 
gard to Spanish background, on the likelihood of being in an 
age discrepant marriage. If the wife is Spanish (Spanish=l) 
and the husband is not (Spanish=0) , the new variable indi- 
cating ethnic heterogamy (MIXSPAN) is coded 1. If the wife 
is not Spanish and husband is Spanish, then the designation 
is MIXSPAN=1 . If both wife and husband are Spanish, then 
MIXSPAN=0 . 

Education was developed from the 22 educational grade 
level indicators listed in the Census. The variable (GRADE) 



-50- 



is entered into the analysis as an interval level measure. A 
recoding of the variable GRADE is performed to ascertain 
discrepancy in the spouses educational levels. 

00 through 10=1 

11 through 14=2 

15 through 19=3 

20 through 22=4 

If the wife's grade and husband's grade are not equal, 
then MIXEDUC equals (1) . However, if a wife's and hus- 
bands' grades are equal then MIXEDUC equals 0. 

The number of times a person has been married is a 
dichotomy, where l=first marriage and 0=remarriage. Since 
the literature indicates that the previous marital history 
of either partner may affect the selection of mate in a 
different age category, the variable FIRSMAR was created. 

Because no accurate determinations could be made of the 
amount of income earned by each wife separately from outside 
as well as household employment and the amount of income the 
women had before marrying their current spouse, income could 
not be used to predict the social class of the women in the 
age discrepant marriages. However, because education levels 
and social class levels are positively associated, education 
levels will be used instead of social class levels. 

Analytical Procedure 

Because marriages are either age discrepant or not age 
discrepant, the dependent variables in this analysis can 
take on only one of two values, resulting in a dichotomous 



-51- 



dependent variable for all definitions of age discrepancy. 
Dichotomous variables introduce many problems into the use 
of the general linear model approach, commonly estimated by 
ordinary least squares (Hanushek and Jackson, 1977) . For 
multiple regression to provide the Best Linear Unbiased 
Estimates (BLUE) , the error terms must be normally distrib- 
uted and independent of the values of the explanatory vari- 
ables. However, in the case of dichotomous dependent vari- 
ables the error terms are, by definition, also restricted to 
take on but two values. In such a situation the error terms 
are clearly not normally distributed (Agresti, 1984) . 

Error terms created by the dichotomous nature of the 
observations are also likely to be correlated with the 
values of the explanatory variables — clearly in violation of 
the ordinary least squares (OLS) assumption. Moreover, in 
these situations (where the dependent variable, and conse- 
quently the error terms, can take on only two values) , the 
assumption of equal variance of the error terms is also 
violated. Use of an (OLS) technique, therefore, would yield 
inefficient estimates and could render tests of significance 
meaningless . 

A resolution to these problems concerning the dis- 
tribution of the error term is the use of a generalized or 
weighted least squares technique. However, this resolution 
does not address the additional problem of inappropriate 
functional form due to the use of a linear solution for 
equations involving dichotomous dependent variables. 



-52- 



Fitting a linear function to such variables could create an 
equation that yields probability estimates that extend 
beyond the possible probability range bound between 0 and 1. 
These observations suggest, therefore, that a non-linear 
function is more reasonable. The appropriate non-linear 
function is that of the general S-curve. Such a sigmoid 
function implies that a change in probability approaches the 
natural limits of 0 and 1 (Harrell, 1980). 

The use of a maximum likelihood logit technique over- 
comes the difficulties presented by both the violation of 
the (OLS) error term assumptions and the inadequate linear 
functional form. Moreover, as multiple regression minimizes 
the distance between the observations and the regression 
line, logistic regression operates to maximize the proba- 
bility of obtaining the observed values, given the set of 
independent variables (Hanushek and Jackson, 1977) . Al- 
though this technique involves a different algorithm than 
the ordinary least squares solution, its use and interpreta- 
tion are similar. 

Through the calculation of a chi-square statistic, the 
logistic procedure provides a test for the statistical sig- 
nificance of the model. Also, the statistical significance 
of the relative effects of the explanatory variables can be 
computed. A proportional effect for each independent vari- 
able with an interpretation analogous to that of the un- 
standardized regression coefficients in ordinary least 



-53- 



squares can also be calculated through the partial deriva- 
tive formula (Hanushek and Jackson, 1977: 188-189). In 
order to perform this function the logistic regression 
coefficients must be transformed into instantaneous rate of 
change coefficients. These coefficients will estimate the 
proportional change in the predicted probability of falling 
into category 1 rather than category 0 of the dependent 
variable for each one-unit change in an independent vari- 
able. In addition, the procedure also controls for the 
influence of all of the independent variables when evaluated 
as a linear effect tangent to the logistic curve at a parti- 
cular point of comparison. The proportional effect of an 
independent variable, Xk, is Pk, where Pk = bk [p(l - p) ] , 
where bk is the logistic regression coefficient, and where 
[p (1 - p) ] is the variance of the dependent variable at 
the instantaneous point of comparison, p (which typically 
represents the mean of the dependent variable) . These 
instantaneous rate of change coefficients describe the rela- 
tive effects of an independent variable as a function of its 
contribution to the slope of the line tangent to the logis- 
tic curve at this single point of comparison. The value of 
the proportional change in the predicted probability of 
Y = 1 will vary according to the value chosen for p. In 
this study Pk will be evaluated at the mean of the dependent 
variable (Dunteman, 1984) . 



CHAPTER SIX 

FINDINGS AND DISCUSSION 



Descriptive Characteristics 

Descriptive information concerning the women in the 
sample, both homogamous and heterogamous , are presented in 
Tables 2 through 5. Tables 6 through 14 provide the logis- 
tic regression analysis for black and white women respec- 
tively and the entire sample. A summary completes the 
chapter and establishes a focus on the likelihood of 
women's participating in age discrepant marriages. 

In Table 2 the overall characteristics for homogamous 
and heterogamous couples are presented, indicating that at 
least 5,038 or 10.8 percent of the couples are age heterog- 
amous, of these 3.6 percent the wife is older, and 7.2 
percent the wife is younger. Women report themselves as 
head of the household in 1,680 of the couples. 

Of the 46,844 couples, 3,385 or 7.2 percent are black 
and 43, 459 or 92.8 are white. One percent of the couples is 
racially heterogamous. Of the 461 racially heterogamous 
couples, the women are married to husbands who were native 
American, Eskimo, etc. Approximately 3.4 percent of the 
couples are of Spanish origin. Additionally, 1.6 percent of 



- 54 - 



-55- 



TABLE 2 

NUMBER OF AGE HOMOGAMOUS AND AGE HETEROGAMOUS : COUPLES, BY 
RACE 



Variable 


Number 


Percent 


Homogamous couples 


41,806 


10.8 


Heterogamous couples 


5,038 


89.2 


Wife older 


1,680 


3.6 


Wife younger 


3,358 


7.2 


Black couples 


3,385 


7.2 


White couples 


43,459 


92.8 


Mixed Race 


461 


1.0 


Spanish Origin 


1,574 


3.4 


Mixed Spanish 


745 


1.6 


Educational Levels 






14 (High school) 






Females 


20,057 


42.8 


Males 


15,454 


33.0 


15 through 22 






Females 


14,674 


31.3 


Males 


17,885 


38.2 


22 only 






Females 


287 


.6 


Males 


1,392 


3.0 


Remarriage 






Married more than once 






Females 


7,748 


16.5 


Males 


7,951 


17.0 



-56- 



the couples have mixed Spanish origin, i.e., at least one 
spouse is Spanish. 

While it appeared that women reached the high school 
level of education (42.8 percent)more than men( 33 percent), 
further review indicates men remained in school to complete 
higher levels of education. Consequently, 32 percent of the 
wives attended college, while 38 percent of the husbands 
attended college. The educational difference is even more 
pronounced at the highest level where 1,392 men or 3 per- 
cent are recorded in contrast to 287 or .6 percent of the 
women. 

Almost 17 percent of the couples were in their second 
or later marriage. For women, 7,748 or 16.5 percent of them 
had been married more than once. Slightly more men in the 
sample, 7,951 or 17 percent, are in their second or later 
marriage. 

The means and standard deviations for all variables 
used in the analysis are provided in Table 3. The mean age 
of the females is 43.65; while the mean age for their 
spouses is 46.45. Thus, on the average, the husbands are 
2.8 years older than their spouses. This represents the 
pattern of "constrained heterogamy," which has been accepted 
as the pattern in today's society. The slight age differ- 
ence is expected and consequently represents the normative 
pattern. 



-57- 



TABLE 3 

MEANS AND STANDARD DEVIATIONS OF VARIABLES USED IN THE 
ANALYSIS, ALL CASES (N=46,844) 



Variables 


Mean 


Standard Deviation 


Wife's Age 


43.65 


15.37 


Husband's Age 


46.45 


15.81 


Wife's Education 


14.12 


2.80 


Husband's Education 


14.33 


3.56 


Percent Worked in 
1979 (among wives) 


45.32% 


0.50 


Wife's Income 


4857.42 


6393.59 


Husband's Income 


18058.40 


13386.66 


Total Family Income 


24317.95 


15130.03 


Percent in First 
Marriage 


83.46% 


0.37 


Percent Spanish 


6.87% 


0.44 


Percent in Racially 
Heterogamous 


0.98% 


0.10 


Percent in Mixed 
Spanish 


1.59% 


0.13 


Percent in 

Educationally 

Heterogamous 


38.54% 


0.49 



-58- 



On the average, women and their husbands had attained a 
high school education. The mean is 14.1 for females and 
14.3 for males. 

Wives' mean annual income is $4,857.42, whereas their 
husband's average annual income is $18, 058. The mean 
total family income is $24,317.95. Family income, as used 
here, refers to all money receipts received by all family 
members during the preceding calendar year from a number of 
sources including wages and salaries, self-employment, pub- 
lic assistance, and Social Security. The data shown in the 
study are for family income, defined here as combined re- 
ceipts of all members of the family during the calendar 
year. Total family income for the heterogamous couples was 
lower than the income of age homogamous couples. 

Describing the Subpopulations 



Black Women 

The data in Table 4 indicate that the mean age of 
black females is approximately 41.97 years. Their spouses 
have a mean age of approximately 45.1 years. Black women who 
are in homogamous marriages have a mean age of 40.9 and 
their spouses have a mean age of 43.4, which represents a 
mean age difference of 2.43. Black women who participate in 
age heterogamous marriages are older than those in age 
homogamous marriages and their spouses are also, on the 
average, older. The mean ages are 47.3 and 54.2 years, 



-59- 



TABLE 4 

CHARACTERISTICS OF BLACK WOMEN IN AGE HOMOGAMOUS AND AGE 
HETEROGAMOUS MARRIAGES 





ALL 

BLACK 

WOMEN 


AGE 

HOMOGAMOUS 

MARRIAGES 


AGE 

HETEROGAMOUS 

MARRIAGES 




N=3385 


N=2837 


N=54 8 


Mean Age 


41.96 


40.93 


47.27 


Mean Age of Spouse 


45.10 


43 . 36 


54 . 15 


Mean Education 


13.40 


13.56 


12.49 


Mean Education 
of Spouse 


12.68 


12.98 


11.14 


Mean Total 

Family Income 


19470.88 


19858.70 


17463.14 


Percent in First 
Marriage 


81.06% 


84.28% 


64.41% 


Percent Employed 
in 1979 


36.40% 


34.58% 


45.80% 


Percent Spanish 


1.18% 


1.34% 


.36% 


Percent in Mixed 
Spanish Marriages 


.99% 


.91% 


1.09% 


Percent in Racially 
Heterogamous 
Marriages 


1.00% 


.88% 


1.64% 


Percent in 

Educationally 

Heterogamous 

Marriages 


39.00% 


38.31% 


42 . 52% 



-60- 



respectively , which represents a mean age difference of 6.9 
years. 

Black females have higher mean educational levels than 
their spouses. All subgroups, on the average, have edu- 
cational grades levels below high school. However, black 
females' participating in age heterogamous marriages are, 
on the average, 1.35 grade levels higher than their spouses. 

The mean family income for black females is lower than 
the mean income for all families, in general. Furthermore, 
a higher percentage of black women participating in age 
discrepant marriages were employed when compared to black 
women participating in age homogamous marriages. The average 
family income for black females participating in age dis- 
crepant marriages is $17,463, and the average family income 
for black females participating in age homogamous marriages 
is $19,858. 

Of all the black females studied, 81 percent were in 
their first marriage. Additionally, 84.3 percent of the 
black females in homogamous marriages had been married only 
once. A substantial percentage (64.4%) of the black women 
in age heterogamous unions (35.6%) have been previously 
married. 

In terms of educational heterogamy, 39 percent of the 
black women have different educational levels than their 
husbands. Of the black females who are participating in age 
homogamous marriages, 38.3 percent are married to men who 
have a different educational levels. Black women in age 



-61- 



discrepant marriages have the highest (42.5) rate of educa- 
tional heterogamy. 

Slightly over 1 percent of the black women in this 
study are of Spanish origin. In addition, about one percent 
of the black women are involved in a marriage in which one 
partner is of Spanish origin. 

Finally, one percent of the black women in the sample 
are in racially heterogamous marriages. The data indicate 
that the black females who are participating in racially 
heterogamous marriages are also more likely to be partici- 
pating in age heterogamous marriages. 

White Women 

Table 5 presents the characteristics of white women in 
age homogamous and age heterogamous marriages. The mean age 
for white females, overall, is 43.8, and the mean age for 
their husbands is 46.6. Thus, the mean age difference is 
2.8, which is standard for couples in the United States. 
The white women in age homogamous couples have mean ages of 
43.3 and their spouses are 45.7, on the average, resulting 
in a mean age difference of 2.4, which is lower than the 
difference for all white women in the study. However, white 
women in age heterogamous marriages have a mean age of 47.8 
for women and 54.3 for their respective spouses; this 
results in a 6.5 mean age difference. 



-62- 



TABLE 5 



CHARACTERISTICS OF WHITE WOMEN IN AGE HOMOGAMOUS AND AGE 
HETEROGAMOUS MARRIAGES 





ALL 

WHITE 

WOMEN 


AGE 

HOMOGAMOUS 

MARRIAGES 


AGE 

HETEROGAMOUS 

MARRIAGES 




N=4 34 59 


N=38969 


N=4490 


Mean Age of Females 


43.78 


43.32 


47.78 


Mean Age of Spouse 


46.55 


45.66 


54.29 


Mean Grade of Females 


14.18 


14.26 


13.46 


Mean Grade of Spouses 


14.45 


14.58 


13.40 


Mean Family Income 


24695.49 


24954.77 


22445.17 


Percent in 

First Marriage 


83.65% 


86.62% 


57.82% 


Percent Worked in 
1979 by Females 


46.02% 


45.53% 


50.22% 


Percent Spanish 


3.53% 


3.34% 


5.19% 


Percent in Racially 
Heterogamous 
Marriages 


0.98% 


0.88% 


1.84% 


Percent in Mixed 
Spanish 
Marriages 


1.64% 


1.52% 


2.65% 


Percent in 

Educationally 

Heterogamous 

Marriage 


38.51% 


37.87% 


44 . 08% 



-63- 



The majority of the white women in the study have at 
least completed high school, and their husbands have com- 
pleted slightly higher grade levels than the females. How- 
ever, white women participating in age discrepant marriages 
have lower levels of education than those participating in 
age homogamous marriage. In addition, white women in such 
marriages achieved slightly higher levels of education than 
their spouses. 

The mean family income of the white couples is $24,695. 
The white age homogamous couples have a slightly higher mean 
family income of $24,955. A lower mean family income of 
$22,445 is presented for white age heterogamous couples. 

Of all the white women 83.4 percent were in first mar- 
riages. Of white women in age homogamous couples, 86.6 
percent were in first marriage. On the other hand, only 
57.8 percent of the white women in age heterogamous couples 
were in first marriages. 

The question of income contribution by women was ad- 
dressed by calculating the percentage of time they worked 
during the preceding year. Based on the data, 46 percent of 
the white females who were employed in 1979. 45.5 percent 
of the white women in age homogamous marriages were in the 
labor force. Even more white women (50.2) who participated 
in age discrepant marriages were in the labor force. 

Three and a half percent of the white women in the 
sample were of Spanish origin. Those white women in age 
homogamous unions represent a slightly lower percentage of 



-64- 



3.34. However, 5.19 percent of the white women who are 
participating in age discrepant marriages are of Spanish 
origins. Moreover, 2.65 percent of the white women in age 
heterogamous unions are also heterogamous in terms of 
Spanish origin. This percentage is higher than for white 
women who are participating in age homogamous marriages. 

Less than one percent of the white women are in raci- 
ally heterogamous marriages. The percentage for white women 
in age homogamous marriages is .8 of one percent. Of the 
white women in age heterogamous marriages, 1.8 percent are 
also in racially heterogamous marriages. 

White women are generally in educationally heterog- 
amous; however, those women participating in age homogamous 
marriages are slightly less educationally heterogamous. Of 
the white women in age heterogamous unions 44.1 percent are 
educationally heterogamous, which is 6.2 higher than those 
in age homogamous marriages. 

Logistic Regression Results 

Predictions of the likelihood of women's participating 
in age discrepant marriages are presented in Tables 6 
through 14. Each model reported regresses a dichotomous 
indicator of heterogamy on several sociodemographic vari- 
ables, (i.e., race, education, ethnic background and times 
married) . Each table contains the following statistical 
information: (1) couple composition (2) sample size, (3) 



-65- 



logistic regression coefficients [B], (4) parameter esti- 
mates (defined as statistically different from zero when the 
attained significance levels for the various statistical 
tests are less than .05, .01 and .001), (5) the propor- 
tional changes in the predicted probability of Y=1 for one- 
unit changes in the independent variables (also referred to 
as instantaneous rate of change coefficients) , (6) mean of 
the dependent variable, (7) intercept, (8) -2 Log likeli- 
hoods for the complete model, and (9) model chi-square. 

Some general comments about the interpretation of the 
logistic regression model are in order. The multiple 
logistic regression model is used to fit a binary dependent 
variable to multiple predictor variables. The 
interprepation of the logistic regression model indicates 
that when the effect of all other variables in the equation 
would otherwise be a 50 percent probability of an age 
heterogamous marriage that a one level increase in the 
independent variable would increase or reduce the likelihood 
of participation by women into a particular type of age 
discrepant marriage. 

The first model, reported in Table 6, presents 
estimates of the likelihood of participation (for all women) 
in any type of age heterogamous marriage. The findings 
indicate that each of the variables has a significant effect 
on the probability that a woman will participate in an age 
discrepant marriage. 



- 66 - 



TABLE 6 

EFFECTS OF SOCIAL DEMOGRAPHIC VARIABLES ON WOMEN'S 
LIKELIHOOD OF BEING IN AN AGE DISCREPANT MARRIAGE: BLACK 
AND WHITE WOMEN, COMBINED (N=46,844) 



Variable 


Logit 

B 


Predicted 

Proportional 

Change 


RACE 


.435*** 


.042 


GRADE 


-.091*** 


-.009 


FIRSMAR 


-1.459*** 


-.140 


SPANISH 


.215* ** 


.021 


MIXRACE 


.586*** 


. 056 


MIXEDUC 


.227*** 


.022 


MIXSPAN 


.304** 


.029 


Mean, dependent 
variable 


.108 




Intercept 


.071 




-2 Log Likelihood 


29,480.22 




Model Chi Square 


2,501.01/7 





* p < .05. 

** p < .01. 

*** p < .001. 



-67- 



Race significantly affects a woman's likelihood of 
participating in an age discrepant marriage. Being black 
increases the probability of age heterogamy by 4.2 percent. 

The model also indicates that being Spanish increases 
women's probability of participating in an age discrepant 
marriage by (2.1 percent). In addition, if only one of the 
spouses, is Spanish, the woman's probability of being in an 
age heterogamous marriage increases by 2.9 percent. As 
hypothesized, women who are heterogamous in other factors 
such as education and race are also more likely to be in 
age discrepant marriages (2.2 and 5.6 percent respectively). 

Support for the hypothesis that women with higher 
levels of education are more likely to be in an age dis- 
crepant marriage was not derived. In fact, the research 
indicates that as women's educational levels increase they 
are less likely to participate in an age discrepant mar- 
riage. 

Another major finding in Table 6 was consistent with 
previous research and the hypothesis. Women who have been 
married only once are less likely to be in an age discrep- 
ant marriage. Being in a first marriage decreases the 
probability by 14 percentage points that a woman would par- 
ticipate in an age discrepant marriage. Others (Vera, 
Berardo, and Berardo, 1985; Atkinson and Glass, 1985; and 
Rawlings, 1978) have also found that wide differences be- 
tween the age of husbands and wives are more common in 
second marriages than in first marriages. The findings 



- 68 - 



reported here also indicate that remarriage increases the 
likelihood of women participating in marriages in which 
they are 10 or more years younger or 5 or more years older. 

Age heterogamy is operationally defined as women's 
participation in marriages in which they are five or more 
years older than their husbands or 10 or more years 
younger than their husbands. Therefore, analysis of the 
entire sample is next divided into these two types. Table 7 
reports the effects of social demographic variables on 
women's likelihood of being in an age discrepant marriage in 
which the wife is 5 or more years older. 

Only the variables of race, educational level and first 
marriage were significant in predicting the likelihood of 
women to participate in marriages in which they were 5 or 
more years older. Increased grade levels and being in a 
first marriage decreases the likelihood that a woman would 
enter into an age discrepant union. Being black increases a 
woman's likelihood of being older than her spouse by 1.2 
percentage points. 

Consistent with the overall sample, as the educational 
levels increase, the likelihood of womens' participating in 
an age discrepant marriage in which they are at least 5 or 
more years older decreases. When the effect of all other 
variables in the equation would otherwise produce a 50 
percent probability of an age heterogamous marriage then a 
one level increase in education decreases the probability by